| 摘要 |
inputting principal factors (dissolved core temperature, dissolved core concentration, charged amount, age, oxygen content and lance height) out of previously manufactured products to an input layer of an artificial nerve network and calculating a weight between the factors for estimating variations of temperature and carbon concentration aimed at the annealing ending point based on the measured dissolved steel temperature and the carbon concentration, estimating the dissolved steel temperature and the carbon concentration during the measurement and determining amount of oxygen to be charged, comparing the estimated value of the nerve network with a manufactured product value to detect an error and revising the weight between the layers to reduce the error, and applying the revised weight to the next manufacture.
|
